> I'm testing replication to the same master slave server. As outlined in
> http://gborg.postgresql.org/project/slony1/genpage.php?howto_basic.
>
> xxid.so does exist in /var/lib/pgsql2/lib/xxid.so.
>
> I do have two instances of postgres on the same box, the other copy is
> located in /var/lib/pgsql/, however it is not running. I think this may be
> the problem however when compiling the new pg source, I installed it to
> the
> path /var/lib/pgsql2/. I can only assume the program slonik somehow,
> rather
> than looking in /var/lib/pgsql2/lib/ it references in /var/lib/pgsql/lib/.

That oughtn't be it.  Slonik tells the database to load the binary from
$libdir, not from a specific path.

> Any other suggestions :)? Has anyone successfully used slony-I with two
> instances on the same machine. This is important as we use failover
> redundancy across our production boxes, and there are up to three
> instances
> of postgres running on different ports at any one time.

Certainly.  On some systems we have 4 databases on 4 different Slony-I
clusters.

You may want to poke around in /var/lib/pgsql2/lib and /var/lib/pgsql/lib
and "force" some difference (other than with the Slony-I stuff where the
statements are a bit hidden from you) and try loading some other
binary-based thing via specifying that it is in '$libdir/whatever.so'

Another thought; take a peek not just at the output of Slonik, but at the
PostgreSQL logs.  You might have to tweak config to make logging more
verbose.  Run the slonik commands to set up the replica, and see exactly
what queries are put into the log.  They should be pretty useful
"tell-tales" of _exactly_ what is going on.